1. Technical Field
Embodiments of the subject matter disclosed herein generally relate to methods and systems and, more particularly, to mechanisms and techniques for computing subsurface parameters variations (e.g., velocity) at desired depths.
2. Discussion of the Background
Marine seismic data acquisition and processing generate a profile (image) of the geophysical structure under the seafloor. While this profile does not provide an accurate location for oil and gas reservoirs, it suggests, to those trained in the field, the presence or absence of them. Thus, providing a high-resolution image of the subsurface is an ongoing process.
Generally, a seismic source is used to generate a seismic signal which propagates into the earth and it is at least partially reflected by various seismic reflectors in the subsurface. The reflected waves are recorded by seismic receivers. The seismic receivers may be located on the ocean bottom, close to the ocean bottom, below a surface of the water, at the surface of the water, on the surface of the earth, or in boreholes in the earth. The recorded seismic data, e.g., travel-time, may be processed to yield information relating to the location of the subsurface reflectors and the physical properties of the subsurface formations, e.g., to generate an image of the subsurface.
One problem when acquiring seismic data is that one or more portions of the medium (e.g., water) above the surveyed subsurface may have variable velocities. This depth velocity variation creates inconsistent travel-times between the seismic sources and the receivers. For example, as a result of the interaction between warm and cold currents when performing marine seismic surveying, the water velocity may vary rapidly, both temporally and spatially. Thus, the velocity variations may be large enough to have a detrimental effect on subsequent data processing. For example, an oil and gas reservoir may be monitored based on the velocity variations produced by the reservoir. If velocity variations introduced by the warm and cold currents, above the reservoir, are stronger than the velocity variations generated by the reservoir itself, the reservoir cannot be monitored or the obtained results are misleading.
Water velocity variations can be related to the water temperature, salinity and depth. As discussed above, the water velocity changes have implications for seismic processing. Water velocity differences may result in dynamic differences between data in the combined datasets, and these changes may affect the data processing, in particular, processes like multiple attenuation, stacking and 3D migration. However, other layers in the substrate may introduce similar variations. For example, for a land survey, the upper layer (weather layer) may also introduce these variations.
There are methods in seismology that allow computing of fine relative velocity variations (dv/v) in the subsurface. In these methods, correlations of noise records are used to reconstruct Green functions betweens pairs of receivers. Under certain hypotheses, it is possible to compute velocity variations for the coda of the correlated signals as described, for example, in Brenguier et al., “Towards forecasting volcanic eruptinos using seismic noise,” Nature Geoscience, Volume: 1, Issue: 2, pages: 126-130, 2008.
In geophysics, ballistic paths of the reflection on the reservoir are preferably used. The arrival-time delay of the corresponding wavelet is computed. If the reservoir properties are changing (e.g., oil or CO2 concentration, water injection, etc.), the velocity field is modified locally and the arrival-times of the wavelets vary. Determining the variation of the wavelet properties (e.g. arrival-times) allows reservoir parameters monitoring.
However, for this method, the waves reflected on the reservoir can be very noisy due to a weak intensity. If this is the case, the Signal-to-Noise Ratio (SNR) can be insufficient for velocity variation monitoring.
The near-surface layer (i.e., the medium just below the surface) faces daily and/or seasonal variations, called spurious variations, due to changes in temperature, humidity, etc. These variations induce near-surface velocity variations (i.e., noise), which can hide the deep velocity tracked variations. If the wavelet delays induced by the near-surface are greater than the delay due to the reservoir parameters variations, it is not possible to accurately monitor the reservoir.
To improve the SNR, a non-rigid matching was proposed. Non-Rigid Matching (NRM) is a method which estimates the change in two-way time (TWT) of geological features between two seismic volumes, possibly acquired at two different times. The change in TWT may, e.g., be due to a change in velocity in the surveyed area, displacement of one or more geological features, or a change in acquisition geometry (4D “acquisition footprint”). The method, a trace-by-trace matching, operates on pairs of collocated traces from the two surveys. For each pair, a unique operator is designed to cause one trace of the pair to better match the other. A smoothness criterion is typically imposed to ensure that the operators are spatially and temporally consistent. This enhances the contrast between the seismic responses related to changes within the reservoir and the areas where changes are due to acquisition artifacts or noise.
Another method, implemented by the assignee (CGGVeritas) of this patent application, consists of burying the receivers and/or the sources. The advantages of this method are (1) a significant decrease in noise level, and (2) a protection against daily/seasonal variations because the direct reflections do not propagate through the near-surface.
Although this last method works well, there are cases where it is not sufficient, in particular, when (1) surface wave energies are too high, and (2) the ghosts (or free-surface reflections) are mixed with the useful signal. In this case, the seasonal/daily variations are present.
Regarding the velocity variations computations in the marine field, significant work has been performed to remove the water-layer velocity variations between two successive acquisitions on a given area, for example, “The impact of water-velocity variations on deepwater seismic data,” The Leading Edge, 2003, U.S. Pat. No. 7,321,526, U.S. Pat. No. 6,799,118.
U.S. Patent Publication No. 2007/0268780 discloses a method for removing move-out computation uncertainties. This method uses a collection of traces with similar offset, azimuth and common-depth-point (CDP).
However, all these methods consider only the water-layer velocity variations' contribution removal. In other words, the existing methods do not consider the contribution removal of other layers, above the targeted depth but below the water-layer. Further, some of the existing methods describe an indirect delay computation or indirect velocity determination for compensating the spurious variations. However, this computation requires first a move-out step. The methods also assume a model (water-bottom depth, water-layer and earth velocity model). The methods further assume slow variations of the water-layer, or use only the water-bottom reflection to correct the computation, and the methods do not take advantage of source and/or receiver arrays. The slowness is not used to compute the incidence angles, and the methods do not consider the case where it is not possible to recover the incidence angles.
A method used in 4D land acquisition, the “cross equalization” technique, is described in, e.g., Ross et al. “Inside the cross-equalisation black box,” The Leading Edge, 1233-1240, 1996. Some improvements to the technique to reduce amplitude bias were introduced by Rickett and Lumley, “A cross equalization processing flow for off-the-shelf 4D seismic data,” 68th Ann. Internat. Mtg. SEG Expanded Abstract, 1998.
The method considers several stacks of the same area acquired at different times. The 4D processing consists in searching time-lapse variations at depth. However, this method has problems due to the static variations occurring at the near-surface, which hide the depth variations. To correct this effect, a reference wavelet in a given window is chosen at a first acquisition (signal s1). For a second acquisition, a control wavelet is chosen in the same window (signal s2). The algorithm computes an operator A so that:As2−s1≈0.
The operator A can be computed in the time or in the frequency domain. In the frequency domain, the following relation is obtained:A(ω)s2(ω)−s1(ω)≈0.
The operator A is supposed to contain the near-surface variations between the times of the two acquisitions. By applying operator A to the whole trace, the algorithm is able to compensate the near-surface variations' effect at depth.
This operation is performed after stack, and thus, it suffers from the approximation due to the normal move-out (NMO) operation. Another drawback is that the same operator A applies for all the traces having the same common mid-point but with different source and receiver points. This last issue is addressed by Meunier et al. “Determining acquisition parameters time-lapse seismic recording,” 59th EAGE conference and Exhibition, 1997. In the method proposed by Meunier, the cross-equalization is applied before NMO, between each source-receiver pair, leading to a surface-consistent correction.
To summarize the deficiencies of the existing methods, it is noted that in the conventional 4D exploration, the time window used to compute the “cross-equalization” correction may contain several mixed wave arrivals with different time-evolving variation; using buried sources is limited by their low power so that the body wave reflected off the reservoir interface (hereafter called useful reflection) suffers from a low SNR, which fails to provide an efficient tracking of velocity variations; and the “cross-equalization” correction is performed after NMO correction and requires a reliable velocity model to track fine velocity variations.
Accordingly, it would be desirable to provide systems and methods that avoid the afore-described problems and drawbacks.